DIY LiFePO4 bank: individual prismatic cell fusing

[aybabtme]

Agree, if you use a BMS per parallel string then it's fine.

[CharlieJ]

Heinz #10:
I do not understand this comment:

Sorry to break it to you, Antoine, but deep inside you might already know that this BMS of yours is not ideally suited to the job. Good luck shoe-horning the rest of your LFP setup around this device (rather than re-selling it while it's new).

And by that I'm not saying this BMS is useless, but it's spec'd for 1-16 cells in series.

I have designed and installed two 10p8s LFP banks on a 60' catamaran, each controlled by an Orion Jr BMS. Works fine.

[Cpt Pat]

Originally Posted by CatNewBee

You cam ommit all 3 and rely on the main battery fuse.

Each individual 4S string (three in all) should have its own fuse. Otherwise, if a cell shorts (a rare but serious occurrence) in one string, without the fuses at the end of each 4S string, fault current will pass uninterrupted to the string with the shorted cell, placing the entire pack (and your boat) in peril. Since each undamaged string (the other two) will pass about half the total current to the "bad" string, "on a good day" only the string with the fault will blow its fuse. Don't count on that, all three fuses could blow from the transient current. But you have saved your pack and eliminated much more severe outcomes (fire).

But 12 fuses? I see no point to either that or to connecting the intermediate cells in the strings in parallel. I don't see where any improvement is gained by doing that, unless you are trying to economize on cell balancers - which won't work properly if any of the cells in each parallel "node" aren't closely matched for capacity... and remain closely matched over time.

[hzcruiser]

Originally Posted by CharlieJ

Heinz #10:
I do not understand this comment:


I have designed and installed two 10p8s LFP banks on a 60' catamaran, each controlled by an Orion Jr BMS. Works fine.

I'm sure it'll work, Charlie, but you're talking about 8S whereas the OP suggested 4S. Big difference. Given that the BMS is designed for up to 16S, you're really saying it is perfectly suited for him, too?

What other features apart from LVC/HVC and balancing are being used that justify the 10x price tag?

[CatNewBee]

Originally Posted by Cpt Pat

Each individual 4S string (three in all) should have its own fuse. Otherwise, if a cell shorts (a rare but serious occurrence) in one string, without the fuses at the end of each 4S string, fault current will pass uninterrupted to the string with the shorted cell, placing the entire pack (and your boat) in peril. Since each undamaged string (the other two) will pass about half the total current to the "bad" string, "on a good day" only the string with the fault will blow its fuse. Don't count on that, all three fuses could blow from the transient current. But you have saved your pack and eliminated much more severe outcomes (fire).

But 12 fuses? I see no point to either that or to connecting the intermediate cells in the strings in parallel. I don't see where any improvement is gained by doing that, unless you are trying to economize on cell balancers - which won't work properly if any of the cells in each parallel "node" aren't closely matched for capacity... and remain closely matched over time.

oh, of course, if you intent to make 3 strings and then parallel them, each string is a separate battery and should have its own breakers, fuses, bms. I was referrinf to one string of four blocks of 3 parallel cells each.

[nebster]

I know you already decided to forgo the fusing, which I think is the right choice as well. But for edification/future readers considering this route:

From experience, fusing on the terminals of a high-current design is super tricky to get right. The terminal fuses have to be pretty heavy duty, and consistent, and they can be tricky to torque correctly and get lined up neatly... moreso if you're using rigid busbars, I suspect. If you get fuses that aren't built to good tolerances, or if you miss a Newton-meter on one spot, you'll induce imbalance. It's easier to do that by accident with more and more interconnections.

Bussing up lithium cells is tedious and somewhat dangerous. I absolutely hate working around mine, moreso in cramped quarters like a lot of us have. It's just really easy to accidentally short something, and adding big fuses in the way can only make that more "exciting."

Finally, the fuses themselves do indeed make lots and lots of heat... and it's not even necessarily that that means too much parasitic loss and too much lost capacity, since you can always just size up your cells/pack a little to account for it. No, the real issue is that they dump all that heat right onto the batteries, which is oftentimes the exact opposite of what we want with lithium chemistries.

So intra-cell fusing of prismatic cells is one of those things that pretty much doesn't win for our application IMHO.

p.s. You could go with many parallel strings; do the economic math, the physics is fine.

[CatNewBee]

As I said, doable, just have some awareness of the problems that may arise some day. you need a good monitoring of the balancing (to identify a weakening cell not accepting the charge) and the cell temperatures (to identify a bad cell with possibly upcomming shortcuts) to early recognise upcomming problems and mitigate them.

[john61ct]

Just a general reminder.

Using the terms "lithium" or "lithium ion" is too general.

Also, many techniques used in high C-rate contexts like EV propulsion are not relevant, even counter-productive wrt sub-C House bank usage.

Many statements made in this thread may be true for non-LFP banks, but not relevant for LFP specifically.

If anyone is working with an LI chemistry other than LFP that should be disclosed specifically, and really, discussed separately.

____
IMO the purpose of using a BMS if you choose to do so, is to monitor each cell individually, which should make per-cell fusing redundant.

Tesla developed its per-cell fusing technique because they are using 100s of those tiny cylindricals rather than large prismatics, are not monitoring individual cells, and want to just keep on operating if a few bad cells get dropped per module.

BTW their "fuses" are just precisely sized wires, in effect per-cell fusible links.

But again, not LFP, and crazy-high C-rates, so it is silly to just blindly consider adopting anything Tesla does, for our House bank use cases.

[CatNewBee]

Battery fuses are good practice even with paralleled FLA. Nothing Tesla specific.